A Virtual Try-On and Size Prediction System Using Image Processing and Web Technologies

This paper presents the design and development of a web-based Virtual Try-On and Size Prediction System capable of estimating clothing size, evaluating fitting accuracy, and generating visual analytics such as heatmaps and confidence charts. The system accepts clothing and person images through three input modes: file upload, image URL, and real-time camera capture using Base64 encoding. It generates predictions using a simulated smart- fit algorithm and provides buyable product links based on clothing type and estimated size. The platform is implemented using the Flask framework and integrates Plotly for data visualization, NumPy for synthetic heatmap generation, and secure file-handling utilities. The system also incorporates user authentication, clothing database mapping, and a unified try-on API, offering a complete end- to-end architecture for apparel recommendation systems. The research demonstrates how lightweight image processing methods can be combined with web technologies to build interactive try-on solutions without complex deep-learning models.